DIY - Dipole Subwoofer

A Gradient SW-63 Look Alike

For a long time I have been thinking about a DIY dipole subwoofer project, but since I have no need for one myself it has not developed any further. Now, when I am launching my crossover project the project is increasingly interesting, but time is limited and I have not been able to make it myself.

Then one day I got an interesting email from Paal Jensen in Norway who was interested in my crossover, he has made a Gradient SW-63 copy and it looks very neat.
Paal's Homepage

Paal Jensens ESL63 with DIY subwoofer  Paal Jensens ESL63 with DIY subwoofer fro the side

Paal Jensen's Drawings

Paal was very cooperative and sent me the drawings he used for making the subwoofers, these are found below:
SW-63 clone drawing (pdf)
 

Some Comments on the Drawings

The text is in Norwegian and even if I think you get what he means, here are some translations:
tykk = thick
stk. = pcs.
ca. = appr.

The construction looks quite straight forward and the most difficult part is to get the curved sides right and an easy solution is to remove the top plate from one of your ESL-63s (push it gently to the right and lift it), and use this plate as a ruler.

Paal has not included the top cover in the drawing, in order to make this you need to modify one of the bigger (ca. 13mm thick) plates and I am sure you can figure out how to do it.

You may need to adjust the holes, depending on the woofer you use.

The woofers shall be mounted pointing in different direction to reduce the distortion.
Opposite direction of woofers
Electrical connection:
Frontal directed speaker - in phase.
Rear directed speaker - in anti-phase.

Best result is obtained if the speakers are connected in parallel, resulting in a 4 ohm impedance with 8 ohm speakers. It is also possible to connect them in series, giving a 16 ohm impedance for two 8 ohm speakers.

For the covers Paal has used stretch fabric.
 

Suitable Woofers

You can't use all elements of a certain size, it has to have a very low resonance frequency, be able to move lots of air, have a high damping factor and a frequency response with a 6dB/octave slope when mounted in free air. Not many elements fulfill these specification. The parameters that controls the behaviour of a woofer in a dipole application are:
Fs (resonance frequency)
Qts (total damping factor)

The resonance frequency shall ideally be below 20Hz for use with full range dipole subwoofers.
The Q-value shall ideally be 0.5-0.8 for use with dipole subwoofers.

Fs can only be altered by mechanical modification of the woofer, but Qts may be altered to suit your needs by application of a calculated series resistor. The formulas used for this are:
Qes'=Qts'*Qms/(Qms-Qts')
Rs = Re*(Qes'-Qes)/Qes

If we for example want to alter the Qts in the Peerless 831857 driver from 0.46 to a more suitable 0.6 the figures are (Qms = 3.90, Re = 5.5):
Qes' = 0.6 x 3.90 / (3.90 - 0.6) = 0.68
Rs = 5.5 x (0.68 - 0.46) / 0.46 = 2.6 ohm

When you apply the series resistor you must subtract the amplifier output impedance and cable resistance.

The Peerless 315 SWR (831857) is indeed suitable for this application and Paal Jensens measurements on his subwoofers (with Peerless 315 SWR) and ESL-63 with a standard XO-2001 crossover (6dB/octave dipole correction) can be seen below.
Paal Jensens SW
The room interaction is included in this graph and the resonance at 45Hz masks the drop at low frequencies, overall it's a pretty good result. The dip at 170Hz is probably caused by cancellation from the rear emission and moving the microphone would change this.

I also received a nearfield (30cm) measurement result from Paal:
Nearfield measurement of dipole sub with Peerless 831857

The ESL and SW curves should mix at 100Hz and this means that the ESL level shown in the graph is too high. Overall the result is very good and shows that the Peerless 315 SWR element is usable without extra corrections. The -3dB point is close to 20Hz and the SW curve looks pretty much as it shall, the high frequency roll off starts a little early but this may well be caused by the measurement technique.

The Peerless 831857 is actually a good choice. The expected frequency curve with 6dB/octave dipole correction is shown below (Fs = 24Hz, Qts = 0.46).
Peerless 831857 in an infinite baffle

With added resistance the Qts is a little higher (0.6 shown) and the expected frequency curve looks like this (and matches Paals measurements):
Peerless 831857 with Qts 0.6 in an infinite baffle

Peerless 831857 data:
Fs = 24 Hz
Qts = 0.46
Xmax = +/- 9mm
Moved air volume = 0.47 liter
Price: $82 (Madisound)
Information on 831857

If you need to get all the way down to 20Hz (-3dB) you need a woofer with lower Fs and here are some examples:

1252DVC:
Fs = 15 Hz
Qts = 0.53
Xmax = +/- 6mm
Moved air volume = 0.3
liter
Price: $48 (Madisound)
Information on 1252DVC

Dayton DVC (295-185):
Fs = 21 Hz
Qts = 0.37
Xmax = +/- 15mm
Moved air volume = 0.76 liter
Price: $109.85 (Parts Express)
Information on Dayton DVC

Sonicraft 12"
Fs = 19 Hz
Qts = 0.456
Xmax = 13mm peak
Moved air volume =
0.66 liter
Price: $125 (Madisound)
Information on Sonicraft12"

Titanic 1200
Fs = 16 Hz
Qts = 0.407
Xmax = 14mm one way (linear)
Moved air volume = 0.65 liter
Price: $149.80 (1-3) $138.90 (4) (Parts Express)
Information on Titanic 1200

NHT 1259
Fs = 16.5 Hz
Qts = 0.445
Xmax = 13mm peak
Moved air volume =
0.64 liter
Price: $150 (Madisound)
Information on NHT1259

Some of the elements above use double voice coils (DVC) and if you use only one voice coil the Qts is approximately two times higher than with both coils connected in parallel. If you apply a resistor in parallel with the not used coil you get more damping and it is possible to tune the system. We need to make some experiments with this connection alternative before we know how it works in practice, but the idea is very interesting.
 


Suitable Woofers (with altered filter characteristics)

Peerless has made a new and very good woofer, the SWR 308 XLS (830500), this is suitable for dipole action only if the filter characteristics is altered. The reason is the very low Qts (0.20) and the expected frequency response with 6dB/octave dipole correction is shown below:
Peerless XLS308 in an infinite baffle

For use in a dipole subwoofer we need to raise the bass even more with this element, not impossible at all. The XO-2001 crossover can quite easily be modified for 12dB/octave correction, but you will need a lot of power to produce 20Hz. Lots of information on this driver can be found on Linkwitz Lab (Woofer3 page).
The Q-value shall ideally be 0.5-0.8 for use with dipole subwoofers.

Peerless SWR 308 XLS (830500)
Fs = 18.1 Hz
Qts = 0.20
Xmax = +/- 13.5mm
Moved air volume = 0.58 liter
Price: $122 (Madisound)
Information on SWR 308 XLS

I tried making a correction network in my simulator and I think this altered XO-2001 filter suits the element:
Frequency response correction for Peerless SWR 308

 

Possible Alterations to the design

Instead of two 12" woofers it is also possible to use one 15" element, the design must then be around 8 cm higher and the Quads may be raised a bit too much for use in some rooms.

You can use four 12" elements mounted as in the example below:



The example is taken from: Johns Audio

Mounting the elements like this is very good from a vibration stand point, but it has one major disadvantage, the depth of the 'box' is deep enough to give a quarter wave resonance in the subwoofer's active region. This can be reduced by a filter like Mr. Linkwitz has done, but it does not solve the problem all the way. The resonance can be as high as 10-12 dB and if it is at 200Hz, the filter can take away the resonance peak at 200Hz but there are also distortion products in the sound. At 100Hz the 2nd harmonic (200Hz) is likely to be 25dB below and this gets 10 dB stronger because of the resonance, not good...

More information on dipole subwoofers is available from:
The Subwoofer DIY Page
Linkwitz Lab
Geert Meddens Audio Page - Dipole Speaker



MT Audio Design Logo


This site uses frames, but if you have found this page via a search engine and do not see any frames, you can use the button below to get to the main frame of the MT Audio Design site.